These values from model predictions were fit against the corresponding experimental data by the non-linear least-squares method, which yielded the best-fit model parameters, including the energy of the folded state and the transition state.
The FI and standard models predictions were fit to the data by minimizing the mean squared error between the theoretically predicted perceived upward motion direction by the model and the physical true 0° upward direction (using the simplex algorithm, cf. Press et al. 1997).
The nonlinear model predictions are fitted with the drag torque and coefficient of restitution previously identified.
Different prediction models were fitted to the experimental resilient modulus, the Mohr Coulomb envelopes were obtained, and the Huurman's model was fitted to the experimental creep data to provide the necessary parameters for eventual numerical simulations.
Prediction models were fitted using data from a factorial design with three factors: inulin (2, 5, 8 g/100 g), protein (3, 4, 5 g/100 g) and calcium (100, 200 mg/L) concentrations, and afterwards were validated with data from a central composite design experiment, where the same factors, but with different levels were evaluated.
Separate prediction models were fitted to estimate percent density and breast size; absolute dense area was obtained from the product of percent density and breast size.
For a simple univariate regression setting, two different prediction intervals were fitted: Both hold the sample coverage, but only one holds the conditional coverage.
Prediction models were fitted based on the training set, using the prior risk, MAP, and one or more of the significant metabolites.
Bagging is a popular method to estimate models with improved prediction performance by reducing the variance of a single weak prediction model by aggregating the predictions of several weak models that were fitted on bootstrap samples.
